Equity and Ethics in Data Amidst the Covid-19 Pandemic: Lessons from Some Connecticut Data Users
On June 10, 2020, a group of Connecticut data creators and users gathered virtually at our Let's Talk Data series to discuss equity and ethics in our data work, especially amidst the Covid-19 pandemic.
This was the first time our community of practice met. A community of practice is a group of people with common work or interests who talk about theory and practice issues to improve their work. Group members each possess different skill levels and knowledge and also come with a common belief that they can learn from and contribute to one another.
While there are codes of ethics for various professions that data users may belong to, there is no unified code of ethics for working on data projects. The Covid-19 pandemic has highlighted some of the challenges around data creation and use: Who gets to create the data? Who can share the data? Who makes the decisions about the distribution of resources based on the data?
We centered part of our discussion around an article about data collection and subsequent planning of the Covid-19 pandemic that resulted in a lack of resources for the hardest-hit communities in Louisiana," This is what happens to us." How U.S. cities lost precious time to protect black residents from the coronavirus.
We focused on some central questions:
How can we identify blind spots in our own data collection?
How can we partner with those we seek to serve in data collection and analysis?
How can we improve our practices to reduce the likelihood of harming the people we seek to help through our data work?
Overarching Lesson: We need to take a broad approach.
We can't take on ethics and equity in silos. Instead, we need to develop an organizational equity and ethics framework that cuts across all data work. Since staff who are not "data staff" make decisions about and communicates data, too, this framework should be organization-wide and apply to all aspects of the organization.
As an example, one group member talked about the racial justice framework their institution utilizes. This framework guides their staff training, policies and practices, and projects. The staff participates in racial justice training related to data as well as other aspects of organizational work. Practically, the data team disaggregates all of its data by race and ethnicity so that the organization can look at where there are positive or negative disparate impacts. They also work to help people understand what the data means and how to use it.
Each of the ideas below would be best considered in light of this overarching framework.
1. Involve the Populations We Work With or Serve.
One important theme that arose from the discussion was the importance of meaningfully engaging individuals that our organizations or institutions work with or serves (terms include "stakeholders" and "consumers").
One person talked about ensuring that stakeholders or consumers sit on the organizational board. To ensure that stakeholder ideas and perspectives are meaningfully considered, board rules, policies, or practices may need to be changed.
Another way to engage with the population we work with is to provide resources and logistical support for stakeholders in projects and initiatives that they create and lead.
Lastly, we discussed having stakeholders involved in the data work itself, including identifying how they do and do not want their data used. Taken further, we can provide stakeholders with training to use their data, explore the questions that matter to them, and work to use the results to improve their own lives and communities. And we can provide support for them in their data work.
2. Identify Data Omissions at The Start.
The article highlighted the problems of data omissions early on in the pandemic in Louisiana. State health officials didn't look for cases unless people with symptoms had contact with people who had recently traveled overseas. Like most states, they did not report data on the racial or ethnic demographics of coronavirus tests. And they weren't releasing census tract or block-level data about cases.
The mayor of Shreveport, Adrian Perkins, had a hunch that there were cases in his city and decided to find out by using data the emergency responders were collecting. He created a physical map with pins representing potential cases and found that some of the Black neighborhoods already had clusters of cases. Yet nobody was expecting or even looking for them. He attempted to alert health officials and seek resources, to no avail. The result was delayed response and lost lives.
One of our group members pointed out that when looking at data project plans, she looks for holes or "errors of omission." What groups have we left out of the data exploration? Even if our data collection does not directly harm people, the intervention intended to help them will potentially harm or fail to help them.
We cannot always know whom we may negatively harm through data exploration. With our framework as our guide, we can build check-ins into our process where we ask ourselves whom we are leaving out, or who we might be harming. The only way to reduce the likelihood of hurting those we seek to help is by asking the question and being honest about the answer. We must be intentional about whom we include, rather than unintentionally excluding racial or ethnic groups or other groups that we may be disparately impacting.
3. Sensitive Data Can Be Hard to Collect, So We Need to Identify Proxies.
Many of us were able to identify "sensitive" data that we would like to collect—information about individuals that people often don't want to share. Many of us have tried to collect data on participant income. It can help ensure that our programs are reaching the people who most need them or apply for additional funding. But income is a question that is either answered inaccurately out of shame or may not be easy to answer due to the nature of people's incomes. Not everyone has an annual salary.
We talked about some proxies to learn information that can help serve people with low incomes or people living in poverty.
One participant uses census tracts as a proxy. If having the address of individuals you serve is essential, you can identify the census tract in which they live. You can identify demographic characteristics of census tracts that are most important for your services, and create a schema of the vital demographic characteristics related to the services you provide.
These are suggestions for proxies we can use, with the caveat that this information must be relevant to the program.
Medicaid benefits (in health care setting)
SNAP benefits
Receiving free or reduced lunch
One person mentioned that they include a note that these questions will not affect any services they will receive from the organization.
4. "How Much Should We Share?"
One participant raised a question connected to the overarching theme of the discussion: how much do we need to know, and how much should we share with others? At a base level, we should only be collecting data that we're going to use. This requires careful planning and should involve the people we are serving to help us learn what could be hurtful or helpful for them in the data inquiries we are considering.
This also relates to what we may typically think of as data ethics: consent. Do we share enough with the people from whom we collect data so they can truly consent? Do we share data about them in a way that can lead people to draw harmful, hurtful, or incomplete conclusions?
Bringing It All Together
This session raised lots of questions for me and gave me some ideas for how I can improve my data work. I look forward to learning with this group as we seek to improve our data practice and respond to the Covid-19 pandemic in ethical and equitable ways.
Connect With Us
You can learn more about the Equity in Data Community of Practice here or look through the resources that have been shared from our sessions. We meet monthly, and you can sign up to join us here (curiosity and interest in data are the only requirements!). If you are interested to learn more about CTData, check out what we do and the services we provide. For training and tips on how to use data to inform your personal and professional life, register for one of our CTData Academy workshops or browse our blog. You can keep up with us by subscribing to the CTData newsletter and following us on Twitter, Facebook, and LinkedIn.